This study investigates energy allocation strategies for Internet of Things (IoT) clusters powered by energy harvesting, addressing the challenge of equitable and efficient energy distribution among nodes. Employing a game-theoretic framework, we compare cooperative (Shapley value-based), non-cooperative (penalty-driven), proportional, and priority-based allocation methods across 30 variations of penalty (20–60%), top allocation percentage (30–70%), and allocation strictness (Strict vs. Permissive). A simulation with 10 nodes over 24 time steps evaluates fairness (Gini coefficient), efficiency, wastage, and unmet demand. Results indicate that cooperative, proportional, and priority methods achieve 0% unmet demand and 100% efficiency, with Gini coefficients ranging from 0.094 to 0.350. Non-cooperative Strict allocations exhibit higher unmet demand (0–50.1%) and lower efficiency (0.499–1.0), improving with higher top percentages and Permissive settings. The optimal variation (Penalty=30%, Top=70%, Permissive) balances fairness (Gini=0.125) and efficiency (1.0). Cooperative strategies are recommended for IoT deployments, with tuned non-cooperative methods as viable alternatives. Future work includes dynamic penalty adjustments, heterogeneous node models, and real-world validation. This research contributes to sustainable IoT systems by providing a robust framework for energy management.